In some architectures, an assembling device has to access many distributed servers in order to obtain a decodable quantity of erasure-coded fragments from a distributed storage system. Commonly used push protocols for accessing the distributed servers may be susceptible to fault conditions and may require a significant setup time. Basic principles and details relating to erasure codes and rateless codes may be found, for example, in the publication “Digital Fountains: A Survey and Look Forward” by Michael Mitzenmacher, Harvard University; and in publications by Michael G. Luby and Digital Fountain Inc.